Abstract The overarching goal of the Characterization Unit is to capture the molecular and cellular heterogeneity and clonal evolution of three major pediatric cancers, high-grade glioma (pHGG), neuroblastoma (NB), and very high-risk acute lymphoblastic leukemia (VHR- ALL). Although many studies have characterized the interplay between DNA methylation, chromatin accessibility and transcriptome using bulk tumor samples, how the epigenome and transcriptome relate at the single-cell level and how such relationships contribute to cancer heterogeneity is not well understood. In addition to factors intrinsic to malignant cells, extrinsic factors in the tumor microenvironment also need to be better characterized at single-cell resolution. With these considerations in mind, we will apply a two-pronged, spatial multi-omics approach in this project. Our approach is enabled by the diverse array of cutting-edge single-cell and in situ technologies developed by our team of investigators, including: massive parallel single nucleus RNA-Seq, single-cell transposome hypersensitivity site sequencing, multiplexed single molecule RNA fluorescence in situ hybridization and imaging mass cytometry. The multidimensional and longitudinal data set will be analyzed and integrated by the Data Analysis Unit to construct atlases of the tumor ecosystems before and after therapy. We propose the following four specific aims: 1) To identify somatic and germline mutations using whole genome sequencing of bulk tumor samples and blood samples. For each cancer type, we will collect tumor samples from 20 patients before and after medical therapy. The mutation data will be integrated with single-cell omics data to construct phylogenetic tree of cancer evolution during therapy; 2) To conduct transcriptome profiling of both bulk and single-cell tumor samples. For VHR-ALL samples, we will profile malignant cells only. For NB and pHGG samples, we will profile malignant cells, immune cells, and stromal cells in the tumor microenvironment; 3) To conduct epigenomic profiling of both bulk and single-cell tumor samples. We will profile 5-methylcytosine and chromatin accessibility using whole-genome bisulfite sequencing and transposome hypersensitive site sequencing, respectively; 4) To conduct targeted in situ RNA and protein assays. A panel of marker genes for each tumor will be nominated based on computational analysis of the omics data collected in Aims 1-3.